Association of prestroke glycemic status with stroke mortality

Abstract

ObjectiveThe role of diabetes as a predictor of mortality after stroke remains uncertain, and there are very few data for pre-diabetes. This study investigated the association of pre-diabetes and diabetes with 30-day and 1-year mortality after ischemic stroke (IS) and primary intracerebral hemorrhage (ICH).Research design and methodsBetween 2006 and 2013, 2076 patients with IS and 586 patients with ICH (median age 79) were admitted to hospital within 24 hours after stroke onset and were treated in a stroke unit, where they underwent measurement of glycated hemoglobin (HbA1c). Diabetes was retrospectively defined based on medical history, diagnosis during hospital stay or HbA1c ≥6.5% (48 mmol/mol). Pre-diabetes was defined as HbA1c of 5.7%–6.4% (39–47 mmol/mol). Stroke severity was measured using the National Institutes of Health Stroke Scale (NIHSS). HRs were used to test the association of pre-diabetes and diabetes with 30-day and 1-year mortality after stroke onset.ResultsAmong patients with IS, 830 had pre-diabetes and 632 had diabetes; 280 died within 30 days and the other 77 within 1 year. Among patients with ICH, 106 had pre-diabetes and 56 had diabetes; 150 died within 30 days and the other 92 within 1 year. In both stroke subtypes, pre-diabetes and diabetes were associated with higher 30-day mortality. In IS, however, the association was limited to patients with prestroke disability and very severe stroke. At NIHSS 25, HR was 1.58 (95% CI 1.07 to 2.35) for pre-diabetes and 1.67 (95% CI 1.14 to 2.46) for diabetes compared with normoglycemia. In ICH, the association was limited to women for pre-diabetes (HR 1.93, 95% CI 1.15 to 3.24) and to men for diabetes (HR 1.78, 95% CI 1.02 to 3.12). Prestroke glycemic status was unrelated to 1-year mortality.ConclusionsBoth pre-diabetes and diabetes predict short-term mortality after acute stroke, but the association varies depending on both prestroke and stroke-related characteristics. These findings may explain the heterogeneous results obtained by previous studies.